A novel RANSAC-based Kalman filter algorithm for object characterization and tracking

Chakravarty, Sumit; Dong, Chaoqun; Wang, Boyu; Banerjee, Madhushri

doi:10.1117/12.2048692

Cited by 1 publication

(1 citation statement)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Among the identificaiton methods, random sample consensus (RANSAC) has been widely used as an effective outlier detection technique for robust estimation. However, because of the large amount of computation required to build the model, much research has been done to implement efficient algorithms, one of which is the combination of RANSAC and the Kalman filter structure [4][5][6][7]. There are some related approaches involving the combination of these concepts, and the aim of these methods is that the fitting process in the RANSAC begins again with an initial value obtained from the Kalman filter prediction and searches for inliers in the confidence interval [4].…”

Section: Introductionmentioning

confidence: 99%

1-Point RANSAC UKF with Inverse Covariance Intersection for Fault Tolerance

Kim

Kang

Song

2020

Sensors

View full text Add to dashboard Cite

The fault tolerance estimation method is proposed to maintain reliable correspondences between sensor data and estimation performance regardless of the number of valid measurements. The proposed method is based on the 1-point random sample consensus (RANSAC) unscented Kalman filter (UKF), and the inverse covariance intersection (ICI)-based data fusion method is added to the update process in the proposed algorithm. To verify the performance of the proposed algorithm, two analyses are performed with respect to the degree of measurement error reduction and accuracy of generated information. In addition, experiments are conducted using the dead reckoning (DR)/global positioning system (GPS) navigation system with a barometric altimeter to confirm the performance of fault tolerance in the altitude. It is confirmed that the proposed algorithm maintains estimation performance when there are not enough valid measurements.

show abstract